Download Taussig Winner - Georgia College

The Impact of a Seller’s eBay Reputation on Price
Author: Ryan Mickey
[email protected]
(404) 226-9022
Affiliation: Georgia College & State University
Faculty Sponsor: Dr. John Swinton
[email protected]
(478) 445-2591
June 21, 2017
Abstract: eBay provides a large amount of data – enough data to investigate how markets
address the problem of asymmetric information. This study analyzes 183 auctions of Apple iPod
Touches on eBay.com to find the impact of a seller’s rating on price. I found that a seller’s
percent of negative rating significantly impacts the price they ultimately receive.
The Impact of a Seller’s eBay Reputation on Price
I.
Introduction
Some markets have the problem of asymmetric information, where one party in the
market possesses more information than the other. For example, a seller in the market for used
cars typically has more information about the cars for sale than a potential buyer. Some sellers
may be honest, but others could be dishonest. The dishonest sellers sometimes drive out the
honest sellers leaving only bad cars for sale. Since buyers have imperfect information, both
honest and dishonest sellers charge the same price for used cars causing the honest seller to leave
the market. Such goods in these failing markets are often called “lemon” goods (Akerlof 1970).
But a new market can potentially counteract the market failure of “lemon” goods.
Companies such as Carfax.com and Consumer Reports try to alleviate the asymmetric problem
by providing information to the consumer for a small fee. What happens if a third party
information broker cannot consistently signal the product quality? A potential remedy to this
asymmetric information problem is to use past reputation of the seller to help provide a clue as to
the current motivation of the seller. Auction websites such as eBay.com provide a modern
example of this problem. EBay attempts to alleviate the asymmetric problem by providing a
mechanism that allows buyers and sellers to rate each other after transactions.
In order to test the effectiveness of eBay’s rating system, I measured the impact a seller’s
eBay rating has on the price that they receive from an online auction at eBay.com. Using data
from eBay auctions of the Apple iPod Touch, I found that a seller’s rating has a significant effect
on the price that sellers receive. Therefore, I conclude that eBay’s rating system is effective in
alleviating the problem of asymmetric information in the market.
1
EBay and other online auction sites act as an auctioneer while the sellers assume most of
the responsibility for the quality of their product. Sellers must describe the product and their
policies such as shipping costs, acceptable methods of payment, and return policies. Sellers are
also responsible for shipping the product. Unlike in most commodity markets, eBay users cannot
gather first-hand information about the product before they bid. Buyers rely solely on the seller
to describe the item accurately and correctly. Therefore, the buyer assumes the risk of not
receiving the item, receiving a damaged item, or receiving an inaccurately described product.
The seller incurs the risk of not receiving payment. Sellers often alleviate this risk by shipping
an item only after they receive payment.
Once a seller places an item on sale, any registered eBay user may bid on the item. EBay
utilizes a modified ascending-bid, second-price English auction to determine which bid is
ultimately successful. This auction type allows bidders to bid up the price of a product until time
expires. In a second-price auction, the highest bidder only pays the amount of the second highest
bidder. The winning bid must pay the amount of the second highest bid plus one additional bid
increment. Bidders see if they are the high bidder but do not see the maximum bids of their
opponents1.
After a transaction, both the buyer and the seller may leave positive, neutral, or negative
feedback and any additional comments that they wish on the eBay website. EBay then compiles
a rating for all to see by subtracting the total number of negative feedback from the total number
of positive feedback. EBay also offers more complex rating information. Buyers can rate sellers
on the following specific aspects of the auction: item as described, communication, shipping
time, and shipping and handling charges. EBay also categorizes feedback by time periods into
1
Bidders sometimes practice a bidding method called “sniping” where they enter their bid at the last second to try to
win the item without driving up its price.
2
feedback from the last month, the last six months, and the last twelve months. Finally, buyers
may also leave any comments about a seller that they wish.
II.
Literature Review
When current information on product quality is available at a low cost, expected quality
measured by reputation is irrelevant, but information on current quality is often very costly or
nonexistent. When information about a product is imperfect and buyers cannot observe quality,
individuals often rely on reputation to determine a product’s quality. Sellers who invest in
reputation receive a price premium above their competitors who do not have a good reputation
(Shapiro 1983).
Consider the article by Landon and Smith (1998). They used data from Bordeaux wines
to estimate the effect of current product quality and reputation on price. Landon and Smith
(1998) utilized two separate models to determine price and quality. The quality model included
two components – individual firm reputation of quality and collective reputation. Their model
for equilibrium price included the current quality of the wine, the expected quality determined by
reputation, group qualities, and taste factors. The authors used taste ratings from the Wine
Spectator, the world’s most circulated wine magazine, to represent current wine quality. The
findings of the quality model represented the expected quality in the price equation. The authors
also introduced variables that represented the government-determined wine regions of each wine
and types of grapes used for each wine. Their findings supported the belief that both current
product quality and product reputation play a statistically significant role in determining price,
but reputation’s impact on price was close to twenty times greater than current quality.
In most markets, reputation is difficult to quantify; this complicates research.
Researchers must be able to quantify reputation in order to use regression models that
3
demonstrate the impact of a seller’s reputation on price.
The emergence of ecommerce,
specifically online auction websites, provides researchers a means to quantify reputation with an
abundance and variety of observations. According to the Economist (2004), eBay, the leading
ecommerce site, totaled twenty-four billion dollars in trade in 2003. As discussed earlier, most
ecommerce websites use a feedback mechanism to quantify reputation. Researchers have used
these feedback mechanisms to answer two main questions.
Does reputation impact the
likelihood of a sale? Does reputation affect the ending price of an item?
Eaton (2008) tried to answer the question of whether reputation affects the likelihood of
a sale. The author hypothesized that product characteristics, information variables, and seller
reputation impacted the probability of a sale. Using 208 auctions of Paul Reed Smith Custom
model electric guitars sold on eBay.com during the fall of 2002, Eaton found that each additional
negative feedback left by buyers decreased the likelihood of a sale by 1%. Further, Eaton found
that recent negative feedback impacts the probability of a sale more than older negative
feedback.
Alm and Melnik (2002) estimated the effect of a seller’s reputation on the ending price of
eBay auctions. The authors gathered 450 observations of the 1999 $5 United States gold coin
from eBay.com during the period from May 19 to June 7 of 2000. The data represented ninetyone distinctive sellers. The authors believed that the price of the 1999 $5 United States gold coin
on eBay depended upon the value of its gold content, the features and policies of the auction,
and, most importantly the seller’s reputation represented by their eBay rating. Alm and Melnik
used two different variables to represent the seller’s reputation. They used the seller’s overall
rating, calculated by eBay as total positive feedback minus total negative feedback along with a
variable that is represented by the total number of negative feedback left by buyers.
4
Alm and Melnik estimated the parameters using the Tobit maximum likelihood
estimation with variable but known cutoffs (Alm and Melnik 2002).
114 of the 450 auctions
received no bids at all because the seller selected a starting price above the willingness of bidders
to pay. The Tobit method corrects the different cutoffs providing unbiased, consistent estimates.
Alm and Melnik found that both of their reputation variables had a positive, statistically
significant impact on the price of the coins, but the impact was smaller than they had expected.
Other studies have found similar results. McDonald and Slawson (2002) found that
reputation and price had a positive relationship in a sample of 460 auctions of limited-edition
Harley-Davidson Barbie dolls on eBay.com. The paper also found that the length of the auction
does not affect price. Houser and Wooders (2000) used a standard two-step generalized least
squares procedure to examine ninety-five auctions of Intel Pentium III 500 processors on eBay
during the fall of 1999.
Their paper found a statistically significant, positive relationship
between positive feedback and ending price and a significant, negative relationship between
negative feedback and ending price. Another eBay study that utilized used car auctions found
that seller feedback impacts the price positively (Andrews and Benzing 2006).
While the studies mentioned above found that a seller’s overall feedback (positive minus
negative) significantly affected the ending price, this finding is not shared by all researchers.
Lucking-Reiley, Bryan, Prasad, and Reeves (2007) used 461 eBay auctions of U.S. Indian Head
pennies minted between 1859 and 1909 to determine the relationship between price and
reputation. The paper first tried to estimate the effect of reputation on price using eBay’s overall
rating score, but the authors found it insignificant. Lucking-Reiley et al. then separated the
rating into total positive feedback and total negative feedback.
Their findings found a
5
statistically significant effect of negative feedback on price, but positive feedback was
statistically insignificant.
III.
Methodology
Data
From eBay.com, I collected 183 observations representing seventy-nine unique sellers for
the Apple iPod Touch on April 12, 2008. EBay serves as a terrific data source because it gives
easy access to all auctions that have ended in the past two weeks by performing a simple search
and selecting an option to see only complete listings. Furthermore, you can search auctions older
than two weeks if you have the item number. On April 13, 2008, I searched “Touch” under the
category Consumer Electronics and the subcategory Apple iPod, mp3 Players. I then clicked the
option to show the completed listings. From each auction of iPod Touches on April 12, 2008, I
manually collected the data on the description of each item, the seller’s policies, and the seller’s
reputation. I only considered auctions that ended in a completed sale. I ignored all auctions that
did not meet the starting bid or the reserve price and all transactions that ended by eBay’s BuyIt-Now Feature2. I also left out auctions of iPod touches that excluded ear phones and a USB 2.0
cable or included additional accessories other than the ear phones and cable. IPod Touches that
had been “jailbroken” were also left out3.
It was important that I collected the data on the next day. EBay keeps the seller’s
feedback in real-time. If I waited too long after the auctions ended, the data I collected would
not reflect the feedback that the bidders saw when they bid on the item. Further, an exogenous
2
EBay has a Buy-It-Now feature that allows sellers to set a price that they are willing to accept right then. Bidders
may simply offer the Buy-It-Now price, and the auction ends. Further, some auctions do not end for one of two
reasons. Sellers set the starting bid, and they sometimes over value their item and set the starting price too high.
Sellers also have an option to set a visible or hidden Reserve Price. If an auction does not reach the set Reserve
Price, the seller is not bound to complete the transaction.
3
Jailbroken is a term referring to a hacked iPod that allows users to upload additional applications that Apple has
not sanctioned. An iPod loses it warranty when it is “jailbroken.”
6
change in trends that might influence the supply or demand for iPod Touches is unlikely to occur
in the course of one day.
I chose the Apple iPod Touch for several reasons. The Apple iPod Touch is a popular
new item that sees many transactions on eBay ensuring that I will find enough observations. It is
important to be able to control for the different characteristics of each iPod Touch being
auctioned. The ideal item to observe would be homogeneous for each auction. Although the
iPod Touches on eBay are not homogeneous, I should be able to control for the differing factors
of iPod Touches. IPod Touches have three different memory storage capacities: 8 gigabyte
(1,750 songs), 16 gigabyte (3,500 songs), and 32 gigabyte (7,000 songs). The larger the memory
typically means a higher price. Further, the condition of iPod Touches on eBay varies. I
introduced a variable to try to control for the differences in condition of each iPod Touch. IPod
Touches have a relatively high value making the effects of each variable easier to see. Buyers
also pay more attention to a seller’s feedback when the price is high because they have more to
lose from a fraudulent seller. Finally, the retail prices of iPod Touches are stable. If the item’s
retail price was volatile, I would have a difficult time controlling for the changes in price.
The average price paid for an iPod Touch on eBay on April 12, 2008 was $270.06 while
the highest price was $500 and the lowest was $202. No buyer should be willing to pay more
than $500 because Apple sells their thirty-two gigabyte iPod Touch for $499. Approximately
36% of the iPod Touches were sealed in the original box while the remaining 64% were used.
Almost half of the iPod Touches stored eight gigabytes of memory; approximately 42% stored
sixteen gigabytes, and the last 8% were the 32 gigabytes version. The average cost of shipping
in the sample was $8.94, and approximately 23% of the auctions included insurance in the
shipping charges. The length of a warranty ranged from 0 to 90 days with a mean of around 24
7
days. Almost half of the auctions accepted payments other than PayPal, and the average number
of bids per auctions was close to 24. The percent of negative feedback varied from 0% to 12%,
and the mean was 0.89%. Table 1 provides summary statistics of the data collected:
Table 1: Summary Statistics of 183 Apple iPod Touches
Mean
Dependent Variable
WINPRICE
Independent Variables
NEW*
SIXTEENGB*
THIRTYTWOGB*
SHIPPING
INSURANCE*
WARRANTY
OTHERPAYMENT*
NUMOFBIDS
%NEGATIVE
Standard Deviation
Maximum Minimum
270.0583 61.77242
500
202
0.360656
0.415301
0.087432
8.936885
0.229508
24.38798
0.497268
23.78142
0.891207
1
1
1
39.98
1
90
1
77
12
0
0
0
0
0
0
0
1
0
0.481508
0.494126
0.283242
5.869944
0.42167
26.36946
0.501364
12.2908
1.751063
*Zero-one Dummy Variable
The Model
Theoretically, eBay’s rating system should help buyers determine the integrity of each
seller.
Buyers may be more willing to bid on products sold by highly rated sellers.
Consequently, sellers with higher overall ratings and fewer negative ratings may receive higher
prices for identical products from sellers with lower overall ratings and more negative ratings
(Alm and Melnik 2002). This may not be the case. Lucking-Reiley et al. (2007) found that a
seller’s rating merely increases the incentive to provide good service. In either case, the rating
system should play a valuable role in increasing the integrity of eBay and improve the
satisfaction of its buyers and sellers. The rating system helps correct the market failure of
imperfect information. Sellers must still compete to keep their customers.
8
In order to find the impact of a seller’s rating, I estimated the following model for Apple
iPod Touches using Ordinary Least Squares:
WINPRICE = β1 + β2 NEW + β3 SIXTEENGB + β4 THIRTYTWOGB
+ Β5 SHIPPING + β6 INSURANCE + β7 WARRANTY
+ β8 OTHERPAYMENT + β9 NUMOFBIDS + β10 %NEGATIVE + έ
(1)
The dependent variable WINPRICE represents the price that the winner of a bid pays to the
seller. The highest bidder must pay the amount of the second highest bid plus one bid increment.
The zero-one dummy variables NEW, SIXTEENGB, and THIRTYTWOGB describe the iPod
Touch of each auction. A new iPod Touch described as still sealed in the original box receives a
one for the variable NEW and zero if the iPod Touch was previously used. A sixteen gigabyte
iPod Touch takes on a value of one for the dummy variable SIXTEENGB and a zero for the
dummy variable THIRTYTWOGB while a thirty-two gigabyte iPod Touch receives a one for the
dummy variable THIRTYTWOGB and a zero for the dummy variable SIXTEENGB. If the iPod
Touch has a storage capacity of eight gigabytes, it receives a zero for both SIXTEENGB and
THIRTYTWOGB. I expect that new iPod Touches, on average, receive a higher price than used
iPod Touches, and the price should increase with increases in the storage capacity.
The variables SHIPPING, INSURANCE, WARRANTY, and OTHERPAYMENT
describe the characteristics of each auction that the seller decides. Alm and Melnik (2002)
believed that bidders take into account the entire cost of a transaction when deciding on their
high bid. Therefore, bidders probably incorporate the shipping costs into their bid decisions.
The variable SHIPPING is the total amount a seller charges for the shipping and handling of a
product. Some shipping costs differ by destination. In these cases, I simply used the same zip
code as the seller to determine the shipping price.
Because bidders integrate their entire
transaction costs, I expect SHIPPING to have a negative impact on price. The zero-one dummy
9
variable INSURANCE is one if a seller includes the cost of shipping insurance in their cost of
shipping and zero otherwise. INSURANCE should have a positive effect on price because
people are willing to pay more when they know that they are protected from their item being lost
or damaged during shipping. WARRANTY represents the number of days each seller allows a
buyer to return the item for a refund or exchange4. The typical numbers of days for refunds are
zero, three, seven, fourteen, thirty, sixty, and ninety. I expect WARRANTY to have a positive
impact on price because buyers have longer protection against a faulty item and fraudulent
sellers.
Sellers also choose the methods of acceptable payments. Most sellers accept PayPal, an
online business that provides a method of online payments5. Some sellers accept other forms of
payments such as personal checks, money orders, and cashier’s checks. I added the zero-one
dummy variable OTHERPAYMENT to control for sellers who accept other payments in addition
to PayPal. An item receives a one if its seller accepts other forms of payments besides PayPal,
and zero otherwise. OTHERPAYMENT should positively impact price because some buyers
may prefer payments other than PayPal. Some buyers do not have a PayPal account while others
may believe that PayPal is insecure.
The variable NUMOFBIDS captures the number of bids an auction has when time
expires. The number of bids an auction receives serves as a means to measure the number of
people interested in buying the product. As the number of people interested in an auction
increases, the willingness to pay of some individuals probably increases as well, giving
NUMOFBIDS a positive effect on price.
4
5
Most are refunds because the seller often possesses only one item.
All the sellers in the sample accepted PayPal.
10
As I mentioned earlier, papers studying eBay used different measurements for reputation.
Alm and Melnik (2002) used both the natural log of a seller’s total negative feedback plus one
and the natural log of a seller’s overall feedback measured by positive feedback minus negative
feedback. Andrews and Benzing (2006) also utilized a seller’s overall feedback as well as the
percentage of feedback that was positive. Houser and Wooders (2006) utilized the following
four variables to measure seller reputation: the number of positive feedback, the number of
neutral feedback, the number of negative feedback, and a dummy variable indicating if the seller
recently changed his user ID or is new to eBay in the last thirty days.
EBay reports a seller’s overall feedback and their percent positive feedback on each
auctioned item’s webpage.
Based on Lucking-Reiley et al. (2007) findings that negative
feedback matters significantly more than positive feedback, I decided to use the percentage of
negative feedback (%NEGATIVE) as the model's measurement for a seller’s reputation. A user
can quickly calculate this value by subtracting the percent of positive feedback of the seller that
eBay reports on every item auction page from 100. The percent of negative feedback and price
should have a negative relationship.
Many of the studies highlighted above contained two or more measurements of a seller’s
reputation, but I believe that this could cause problems such as correlation between the two
feedback variables. Further, Lucking-Reiley et al. (2007) hypothesized that a users on eBay
believe people are inherently good until they prove otherwise. Sellers start with a clean slate in
buyer’s minds and are only penalized when they fail to meet a buyer’s expectations. This would
mean to eBay’s feedback mechanism does not allow users to build reputation, but sellers begin
with a good reputation until their behavior changes it. Therefore, the feedback mechanism
11
serves as an incentive for sellers to remain honest and provide the best possible ecommerce
experience.
My model and data are not without drawbacks. First, I did not include any explanatory
variables that described the attractiveness of an item’s site. Some sellers are better at web-design
and can make their pages more pleasing to the eye and easier to navigate. This would be
impossible to quantify and would require that I use my own judgment. I also did not verify the
condition of the iPod Touch that the sellers claimed. This should not cause any problems because
the feedback mechanism should give sellers an incentive to be honest.
Alm and Melnik (2002) pointed out several problems with the use of seller ratings.
First, buyers have little incentive to leave feedback especially if a transaction went smoothly.
Further, bad sellers could potentially change their online identities. EBay recently tried to fix
this problem by making it more difficult for sellers to change their identities.
EBay now
requires them to provide credit card information. Sellers also have incentive to attempt to
manipulate their ratings by bidding on their own items or selling low valued goods to build up
their rating and then selling a high valued item fraudulently. Finally, honest mistakes sometimes
happen. EBay users may respond to comments and feedback, but it then becomes a “he said, she
said” situation. Users often cannot determine who is telling the truth.
IV.
Results
The regression results appear in Table 2. Overall, the model was very effective. Five of
the nine variables were significant at the 99% confidence level while two of the remaining four
variables were significant at the 90% confidence level. The regression had an adjusted R2 of
0.918704 meaning that the independent variables explain just over 91% of the variability of the
winning price around its mean. Using analysis of variance, the model, with a 99% confidence
12
level, rejects the null hypothesis that the explanatory variables have no effect on the winning
price and accepts the alternative that the explanatory variables collectively affect winning price
significantly.
Table 2. Regression Results
Dependent Variable: WINPRICE
Method: Least Squares
Variable
Coefficient t-Statistic
Constant**
211.9263
43.33426
NEW**
34.95416
8.741912
SIXTEENGB**
64.17136
22.51417
THIRTYTWOGB**
188.8704
34.75645
SHIPPING**
-0.820124 -2.838904
INSURANCE*
9.006843
2.455112
WARRANTY
0.081953
1.097022
OTHERPAYMENT*
8.036771
2.159303
NUMOFBIDS
0.148179
1.274826
%NEGATIVE**
-2.126228 -2.684447
R2
0.922724
Durbin-Watson 2.174947
Adjusted R² 0.918704
Log likelihood -779.4832
S. E. of reg. 17.61287
F-statistic
229.5247
S. S. R.
53666.9
Prob(F-stat)
0
* Significant at the .05 Type I error level
** Significant at the .01 Type I error level
The three variables that described each iPod Touch were all significant at the 99%
confidence level and, as expected, had a positive relationship with price. Holding everything
else constant, an auction of an iPod Touch that is new and still sealed in the original box (NEW)
received, on average, $34.95 more than a used iPod Touch auctioned on eBay. On average, a
sixteen gigabyte iPod Touch (SIXTEENGB) had a $64.14 price premium over an eight gigabyte
iPod Touch while a thirty-two gigabyte iPod Touch (THIRTYTWOGB) earned a $188.87 more
than an a eight gigabyte iPod Touch, ceteris paribus.
13
The next four variables describe elements of the auction that sellers set. The cost of
shipping (SHIPPING) negatively affected price with a confidence level of 99%; its coefficient
was -0.820214 meaning that for every dollar the shipping charges increased the winning price
fell by approximately $0.82. With a 95% confidence level, an auction that included insurance in
the shipping costs (INSURANCE) increased the dependent variable by slightly over $9.00.
Although the length of a warranty (WARRANTY) is statistically insignificant, it does have the
positive relationship with price that I expected. The variable OTHERPAYMENT that described
auctions that accepted other payments in addition to PayPal was positively related to the winning
price with a 95% confidence level. The winning price increased by roughly $8.00 if the seller
accepted other payments in addition to PayPal.
The number of bids (NUMOFBIDS) was statistically insignificant, but it did have the
expected positive relationship with the winning price.
I expected the number of bids to
significantly increase price. EBay’s policies probably impact the significance of this variable.
Sellers decide the starting bid price. Most sellers set the starting bid low to attract more
attention, but some sellers set the starting bid high because they want to make sure that their
benefits are greater than their costs. The auctions with a low starting bid typically receive many
bids while the auctions with a high starting bid receive very few bids even though both end at
roughly the same amount. This most likely diminishes the direct impact of the number of bids
on the final selling price.
The most variable of interest measures a seller’s reputation. The percentage of negative
feedback of a seller (%NEGATIVE) was inversely related to the winning price and statistically
significant with a 99% confidence interval. Holding everything else constant, the price a seller
14
received decreases by almost $2.13 for a one unit increase in the seller’s percentage of negative
feedback.
The retail prices on Apple.com of an eight gigabyte, sixteen gigabyte, and thirty-two
gigabyte Apple iPod Touch are $299, $399, and $499 respectively. My model can make some
predictions about the cost of iPod Touches on eBay. A new, thirty-two gigabyte iPod Touch
costs $432.05 on eBay when calculated by using the mean values of the explanatory variables.
When shipping insurance and other payments are included, the cost jumps to $449.09. Using the
seller’s most advantageous combination of variables, a new thirty-two gigabyte iPod Touch with
free shipping and insurance, a 90-day warranty, acceptance of payments in addition to PayPal, 77
bids, and no negative feedback sells for $471.58 on eBay.
V. Conclusion
This paper found that a seller’s rating, particularly negative rating, impacts the price they
receive on ecommerce auction. Since bidders cannot physically examine items up for auction,
they must rely on sellers to provide relevant, accurate information. Sellers may misrepresent an
item or fail to deliver an item once payment is received, and buyers have no way of knowing
which sellers are dishonest. To combat this asymmetric information problem, online auction
sites such as eBay.com developed feedback mechanisms that allow buyers and sellers to rate
each other.
I found that negative feedback tends to decrease the price that a seller receives. This is
consistent with Lucking-Reiley et al. (2007) who also found that negative feedback decreases the
price a seller receives for an item. One possible explanation is that eBay users believe people are
inherently good (Lucking-Reiley et al. 2007). Sellers can only be hurt by their own actions.
This theory does not take away from the value of the feedback mechanism. As long as seller’s
15
future benefits outweigh their potential short-run benefits of cheating, the feedback mechanism
provides sellers with an incentive to describe their items accurately, provide good service, and to
fulfill their obligations.
The rules and laws of ecommerce are changing quickly. According to a recent New
York Times article, eBay recently altered the way it displays search results. EBay originally
displayed the results in order of when auctions ended, but now sellers with better feedback,
lower shipping charges, and lower prices show up first (Stone 2008). Although eBay’s feedback
system is not perfect, this study demonstrates the value of buyer feedback in terms of the impact
a poor reputation has on selling price. If eBay wanted to take full advantage of its feedback
system could introduce an incentive for buyers to record their feedback for every transaction.
Perhaps buyer discounts would increase the frequency with which customers make use of the
feedback fields. This would improve the measurement of sellers’ reputations and ultimately,
improve the quality of both consumers’ and sellers’ eBay experiences.
16
Bibliography
Akerlof, George A. 1970. The Market for "Lemons": Quality Uncertainty and the Market
Mechanism. The Quarterly Journal of Economics 84(3): 488-500.
Alm, James, and Mikhail I. Melnik. 2002. Does a Seller's Ecommerce Reputation Matter?
Evidence from eBay Auctions. The Journal of Industrial Economics 50(3): 337-349.
Andrews, Thomas, and Cynthia Benzing. 2006. The Determinants of Price in Internet
Auctions of Used Cars. Atlantic Economic Journal 35: 43-57.
Eaton, David H. 2008. The Impact of Reputation, Timing, and Source on Auction
Outcomes. The B.E. Journal of Economic Analysis and Policy
http://www.bepress.com/bejeap/vol7/iss1/art33 (Accessed 9 March 2007).
7(1).
A Perfect Market. 2004. Economist 371(8375): 3-5.
Houser, Daniel, and John Wooders. 2006. Reputation in Auctions: Theory and Evidence
from eBay. Journal of Economics and Management Strategy 15(2): 353-369.
Landon, Stuart, and Constance C. Smith. 1998. Quality Expectations, Reputation, and
Price. Southern Economic Journal 64(3): 628-647.
Lucking-Reiley, David, Doug Bryan, Naghi Prasad, and Daniel Reeves. 2007. Pennies from
eBay: The Determinants of Price in Online Auction. Journal of Industrial Economics
55(2): 223-233.
McDonald, Cynthia G., and V. Carlos Slawson. 2002. Reputation in an Internet Auction
Market. Economic Inquiry 40(4): 633-650.
Shapiro, Carl. 1986. Premiums for High Quality Products as Returns to Reputations.
The Quarterly Journal of Economics 98(4): 659-680.
Stone, Brad. 2008. EBay Chief Says Change Isn't Over. New York Times. (April 17, sec. C).
17